find_batch_size#

bayesflow.utils.find_batch_size(sample, memory_budget: str = 'auto', min: int = 4, max: int = 1024) → int[source]#

Returns an estimation of an optimal batch size based on memory budget and sample memory footprint.

Parameters:

  • sample – Any nested structure of tensors, representing a single sample.

  • memory_budget – The maximum available memory for a single batch.

  • min – The minimum batch size.

  • max – The maximum batch size.